1. Introduction

Treelines of High Asia mark the boundary of one of the largest conversions of mountain forests into pastures worldwide. The Qinghai-Tibet Plateau (QTP) has both a drought-line, which contains over 20 tree species and extends towards the deserts of Central Asia, and an alpine treeline, which is one of the highest treelines in Eurasia and one of the most extended mountain forest borders in the world. The treelines of the QTP have been influenced by climate, uplift, and human activity (Miehe et al., 2014; Su et al., 2019; Mao et al., 2021; Wang et al., 2022). However, understanding when and how much these factors have impacted the treelines of the QTP remains a major challenge. This study focuses on how human activity has influenced treeline ecotones in the QTP, yet relevant and latest teeline researches in the region were also involved (Liang et al., 2011, 2016; Körner 2012; Lu et al., 2023).

To understand the extent and timing of human activity on forest cover requires dating the onset of the Anthropocene. The concept of the ‘Anthropocene’, introduced first in 1980 by Crutzen and Stoermer (2000), proposed a new geologic era starting with the emissions of the greenhouse gases of the Industrial Revolution. All further attempts to push back the onset of the Anthropocene relied on regional changes: A second attempt introduced several stages of impact of cultivation and herding as aftermath of the Neolithic Revolution (Ruddiman et al., 2008; Certini and Scalenghe, 2011; Fuller et al., 2011; Smith and Zeder, 2011; Vigne, 2011; Glikson, 2013; Heitkamp et al., 2014; Winiwarter and Bork, 2014). A third and indirect impact is the domino-effect of changed vegetation dynamics following the extinction of predator-naïve mega-herbivores of islands as well as the Americas south of Beringia (Martin, 1984; Doughty et al., 2010; Sandom et al., 2014; Boivin et al., 2016; Malhi et al., 2016). The fourth impact, similarly changing the vegetation structure, is at least partly connected with the ‘Blitz’ - the fire impact of humans, clearly traceable with their arrival on uninhabited islands (e.g., Kershaw, 1986; Ogden et al., 1998) and possibly present in the African savannahs (Dechamps, 1984; Mercader et al., 2021) since humans first utilized fire (Wrangham, 2017). Here, we define the Anthropocene as the age when humans shaped the vegetation cover enduringly, e.g., changing forests into grasslands. This definition results in accurate, region-specific dates for the onset of Anthropocene. Although forest cover changes of the current and past are widely documented, the early human footprint during the Holocene remains obscure (Chen et al., 2015; Miehe et al., 2021).

Few studies have synthesized recent research on treeline changes on the QTP during the Anthropocene. On the QTP, forests are converted to pastures that are referred to as ‘alpine meadows’, however, this is somewhat of a misnomer. ‘Meadow’ is generally defined as grassland managed by mowing and ‘alpine’ as elevations above the natural upper treeline. Regardless, the term ‘alpine meadow’ represents, in a nutshell, the whole complexity of the human footprint in the Tibetan Highlands Anthropocene. The objectives of this paper are to present (1) palaeo-environmental records of Holocene forest changes in the QTP, and (2) current records of the treeline ecotone. This paper summarizes the authors own field evidence over the last 45 years and evidence from recent studies in order to identify gaps in our knowledge. As the focus of this contribution is on the sub-recent and Holocene human footprint, it is advisable first to mention those phenomena which are encountered as natural.

Table 1 Current uppermost tree records across the Tibetan highlands.

2. Basic structures

Forests consist of trees, upright woody plants ‘taller than a person’ (Körner, 2012) with a single or several stems branching from the ground from a wooden base (‘lignotuber’, ‘xylopod’). Trees with a single trunk prevail in nature and perception, but in the treeline ecotone, towards the ‘tree species line’ (Fig. 1) or after disturbance (fire, cutting), some tree species may resprout with several stems. A multitude of local impacts induce a great number of forest-structures, yet a common character of treelines in our area is the decrease in height of trees towards their upper limit. Meanwhile, recent studies on Tibetan junipers suggest that water use efficiency increases, due to enhanced photosynthetic capacity, toward this upper limit (Tao et al., 2021). Fig. 1 summarizes impacts (A-G) why the upper margin of the forest belt, conventionally ‘tree-line’, is predominantly not a line, but a transition zone (‘ecotone’). Almost all treelines are in the reach of humans while untouched treelines are the exception, yet still exist in remote areas. The perception of the treeline's remoteness of human habitation in almost the entire High Asia, thus suggesting its naturalness, is misleading (although there are likely natural treelines in some remote areas where no human settlement exists), as forest fragmentation with its tree-limits is a common site where seasonal summer-grazing settlements were established.

One key characteristic of treelines is the paucity or even absence of young trees, including both ‘germlings’ (i.e., first year of germination) and ‘seedlings’ (i.e., first year after germination), as well as ‘saplings’ (young trees) (Körner, 2012). Many forests in the treeline ecotone consist of mature and old trees hundreds of years in age. The longevity of trees in the treeline ecotone suggests that although in some cases the ground under trees may be covered with seeds (e.g., juniper forest in Reting, north of Lhasa), centuries can pass without recruitment (Körner, 2012). An additional explanation for the absence of regeneration is the impact of livestock, which may trample and browse young trees. For example, the western and northern juniper-communities on the QTP lack rejuvenation; however, in the valleys of the outer eastern highlands, where grazing pressure is lower, the Picea and Abies forests show clear reforestation. These observations are consistent with reports from Iran, where junipers have failed to return after millenia of pastoralism (Noroozi and Körner, 2018). Furthermore, patterns of regeneration (or lack of regeneration) appear to be species-specific. For example, rejuvenation of oaks in treelines is poor or absent. Although climate warming may affect rejuvenation, treelines generally react slowly to climate changes (Liang et al., 2011) and interactions between trees and shrubs may play a more important role (Liang et al., 2016).

Among the wide range of treeline ecotone structure types (Bader et al., 2021), undisturbed treeline-sites to detect the natural structure are rare in the Hengduan Mountains and the entire of High Asia. At higher elevations, the most common treeline ecotone pattern is the change from an almost closed canopy cover to an increasingly open woodland, which gradually ends with irregularly scattered and stunted trees at the outposts. In the humid southeastern Abies forests, isolated trees tower over a closed layer of evergreen broad-leaved Rhododendron thickets, which decreases in height from 2 to 0.5 m but maintains complete cover, additionally with a thick bryophyte carpet. This decrease in canopy cover at higher elevations is also observed in the Juniperus forests of the southern exposures. The trees here do not emerge from closed shrub- and moss-cover, but from bare open soil. However, prostrate Juniperus pingii var. wilsonii in the arid southern highlands, and Juniperus pseudosabina in western High Asia may patchwise form thickets up to three meters, with emerging juniper-trees (e.g., Juniperus semiglobosa).

In contrast to the commonly disintegrating forest structure in the treeline ecotone of conifers, a treeline feature of the southeastern periphery is the hedge-like limit of closed evergreen dwarf–forest. This is observed in the Quercus aquifolioides and Q. pannosa (Fig. 2a) east of Litang (western Sichuan). In the Baima Mountains (northeastern Yunnan), Q. pannosa grow at elevations up to 4270 m a.s.l (Yang et al., 2020), resembling Nothofagus treelines in New Zealand.

A major character of the Eastern Highlands and its River Gorges (Hengduan Mountains) is the clear divide between forest-covered slopes of the River Gorges and treeless peneplains of ‘alpine’ attribution. A large number of forest sites have their upper limit between 4000 and 4300 m a.s.l. at this knick-point between the slope and the peneplain. This holds true for oak-forests of southern exposures as well as the spruce- and fir-forests of the northern exposures; their upper limits are conspicuously often at the edge of the slope against the plateau of the peneplain (Fig. 2b). In contrast to slope treelines that have recently moved upslope, the ecotone between the slope-bound forest and the Kobresia-pastures of the plateau is free from small trees, although open sites, where yak have opened the Kobresia root-mat are available for tree seedlings. The question is whether level sites are unsuitable for trees. The unsuitability of these sites is supported by the observation that Picea thrive on northerly-exposed slopes at 4600 m, but have yet to be recorded on the open peneplains at only 4000 m. A similar case is known from the southeastern oak-ecotone, which has an upper limit between 4300 and 4600 m. Similarly, the Juniperus tibetica forest of the southern highlands grow on sunny slopes up to 4700–4900 m, but are absent in the plains at 4400–4600 m further north. This treeline pattern is one of the enigmas of the highlands.

Another conspicuous case of contrasting vegetation cover between sunny and shady slopes on the QTP occurs on the loess-covered slopes of the northeastern highlands around Anyemaqen. Here, sunny slopes with sparsely covered forb-rich pastures harbor open Juniperus przewalskii forests, whereas shady slopes harbor Kobresia pastures (Fig. 3a). The absence of Betula on the shady slope is a conundrum, as it does not match with low rainfall. In addition, the causes of this treeline pattern remain unclear. It is unlikely that this treeline pattern was caused by grazing, as even heavy grazing leaves ungrazed relics in inaccessible sites. It is also unclear why juniper trees are absent in the dense, felty Kobresia root-mat (Fig. 3b). One possibility is that junipers that re-migrated from their Ice Age refugia to the plateau periphery in the Holocene were unable to establish in grassland (Zhang et al., 2005).

3. Drought-lines and potential areas forests ever existed

The aridity-limits of tree-growth, the ‘drought-line’ of forests in High Asia was not given much attention although the gradients of rainfall are well known. The main area of the drought-line ecotone are the westernmost Kunlun Mountain (Picea schrenkiana, Juniperus turkestanica and J. semiglobosa), the northern and southwestern foothills of the Qilian Mountain (Picea crassifolia and J. przewalskii), the southeastern Dry River Gorges (Platycladus orientalis, Cupressus duxlouxiana and Quercus cocciferioides), and in the southern highlands the upper arid catchments of the Yarlung Zangbo River, Sutlej and Indus (J. tibetica, J. convallium and J. semiglobosa). In general, the so far known humidity thresholds of tree-growth range are between the range of 250 and 130 mm annual precipitation (mm/a): Pinus eldarica (syn P. brutia var. eldarica, Eiljar-Ougi, 200–400 m, 130–150 mm/a in western Aserbeidshan, Schmidt, 2002), and the Central Asian Ulmus pumila (150 mm/a in the Mongolian Gobi; Wesche et al., 2011) seem the most drought tolerant tree species in Eurasia. Another constituent of the drought-line in Central Asia is Pistacia vera in northern Afghanistan, with a limit around 200 mm/a (Freitag, 1972). The conclusion of the most concise treatments of drought-lines of Freitag (1972) in southeast Spain, and of Henning (1975) in the La Sal Mountains of southwestern North America is 200 mm annual rainfall. This threshold matches quite well with the drought-lines in our area (Miehe et al., 2008, 2009a; He et al., 2020). Fig. 4 gives the 200 mm annual precipitation threshold in High Asia. It is worth to note that the areas around Lhasa or Xigaze receive more than 200 mm annual rainfall and would therefore be naturally forested.

Concerning the alpine treeline, the elevational data of the uppermost records of treeline species constituents offer a wide range, and especially the upper limits in Chen (1987) or the Flora of China (Wu and Raven, 1994), usually are several hundreds of meters lower than our (GM) records or the sites shown on satellite images of Google Earth and similar sources (Wang et al., 2022). Moreover, and as mentioned above, there is an almost un-recognized contradiction of uppermost treelines on slopes on the one hand and on the other hand treeless flat terrain with dwarf-shrublands or alpine grasslands several hundreds of meters lower than the highest tree-outposts on slopes, both of southern and northern exposure. Fig. 4b depicts in blue the area not suitable for tree-growth, based on the concept of Körner (2012). The data used here indicate the surroundings of Nagqu or parts of the Alpine Steppe of the central highlands as suitable for forest. However, so far there is no field evidence at all to support this.

Fig. 5 gives the current forest borders in High Asia, east of 76°E with isolated records of trees and forests. The gap between the drought-line of the forest (Fig. 4a in yellow line) and the heat deficiency line of the alpine treeline (Fig. 4b in blue) show the ‘alpine meadows’ as a plagioclimax, a secondary grassland after the clearing of forests. This is analogous to the Altiplano in the Andes, where Polylepis woodlands are confined to rocky slopes and absent from the grazed flats (which are regularly burnt).

4. Early impacts during the Holocene

The timing and intensity of the human impact on treelines in High Asia is poorly understood. On the one hand, the density of archaeological records of early human presence are increasing and meanwhile even the presence of middle Pleistocene Denisovans can be expected (Chen et al., 2020), but their environmental impact remains ambiguous. The three common palaeo-ecological approaches to detect pre-historic human-caused environmental changes, (1) non a-biotic vegetation structures like isolated trees in the grassland, (2) charcoal, and (3) human-indicator sporo-pollen, are poorly perceived or simply ignored.

The question of the human impact on the forest distribution is one of the bitterest controversial issues in environmental research and vegetation ecology, and specifically the perception of current vegetation structures as a possible ancient heritage of human impact is disputed since generations. Therefore the help of an admittedly very simple analogy seems justified: It is accepted that people are normally equipped with a full set of teeth; this normal state however changes with time, and so the last tooth will bear witness to a previously complete set of teeth. Climatic conditions under which trees may thrive (‘forest climates’) are, with marginal error bars, a textbook consensus, and it is a common understanding that every habitat, disturbances excluded, carries its optimal phytomass. The conclusion thus would be, that, provided it is accepted that forests consist of trees, a single tree signifies the former presence of forests (‘Lonely Tooth Hypothesis’; Miehe et al., 2014). A solitary tree in a normal site – extrazonal favourable habitats like water-surplus sites excluded – in a widely treeless environment would thus testify to a once existing forest environment, be it with or without interlocking crowns. If trees are fruiting and seedlings thrive well, the current conditions are suitable for the recovery; if not, the mature trees signify a once better environment.

A key witness to demonstrate this approach, is a Juniperus tibetica tree of 3 m height on a southeast facing ridge, fully exposed to the daily strong valley-up winds at 4350 m, in the completely treeless and desertic western Yarlung Zangbo Valley, 550 km west of the current forest limit (Fig. 6). This pattern of isolated trees in normal sites surrounded by dwarfish vegetation of ‘alpine’ attribution is well-known worldwide, like in East African mountains with Erica trees in ‘afroalpine’ heathlands (Miehe and Miehe, 1994), in the highlands of New Guinea (Löffler, 1979) and the Andes with tree-groves surrounded by tropical-alpine ‘paramo’ grassland. The dispute over the implications of solitary trees in treeless environments believed to be unsuitable for forests, goes back to the controversy concerning the natural forest vegetation in Central Europe before and after the Neolithic landnam event (Gradmann, 1898 vs Firbas, 1949), continued in the Andes as the ‘Polylepis-problem’ (Troll, 1959; Walter and Medina, 1969; Ellenberg, 1979; Laegaard, 1992; Kessler, 1995, 2002; Wesche et al., 2008), flared-up again with the emergence of the megaherbivore theory (Vera, 2000 vs Birks, 2005), and is seemingly virulent now in the Tibetan highlands (Song et al., 2004; Zhang et al., 2005; Meng et al., 2007; Ni and Herzschuh, 2011; Miehe et al., 2014). The problem with reconstructing the presumed natural vegetation from the remnants of the original vegetation found in cultural landscapes with all their unknowns was approached by Freitag (1972) in Afghanistan. He noted that religiously protected islands of forest standing isolated in common grazing land, which is otherwise bare of trees, are indicative of the primeval vegetation. The same applies to church forests in the Ethiopian Highlands or graveyard forests in the Maghreb countries. Similarly, Miyawaki (1998) charted the potentially natural vegetation from the tree stands of temple forests in Japan, and Hilbig (1995) concluded from the occurrence of isolated Larix, Picea and Ulmus pumila trees near monasteries that there were once extensive forests in the meadow steppes of Mongolia. Our acceptance of a now defunct forest belt of Cupressus and Juniperus in the southern highlands and the Dry River Gorges of the Hengduan Mountain from the remains of cypress or juniper trees in normal sites also relies on sacred trees. An inventory of forest relics in the ‘alpine meadows’ of the eastern and southern highlands (Miehe et al., 2008) revealed tree records up to 650 km west of the present forest limit and up to 4900 m high (see red dots in Fig. 5). The search for witnesses of the natural forest cover is even more difficult if the tree, unlike the junipers, is a livestock's favourite browse, like the highlands' birches (Fig. 5). The space between the current forest border (green band in Fig. 5) as given in the Atlas of the Tibetan Plateau (Editorial Committee of the Atlas of the Tibetan Plateau, 1990) and the potential natural treeline (the ‘6 ℃/90 days line’, Körner, 2021) would designate the area where humans converted forests into pastures (Fig. 7).

Charcoal can also be used to detect pre-historic human-caused environmental changes. Note that charcoal is not an incontrovertible sign of human impact, but it helps narrow the possible explanations for certain environmental changes. For instance, one way to determine whether the fire record is cultural or natural is to look for the presence of fire-adapted plant traits (‘pyrophytes’). If pyrophytes are absent, fire is not a driver of ecosystem change on an evolutionary time scale. In the Tibetan flora, pyrophytes have not been observed. In addition, lightning in Tibet is commonly followed by torrential rains. These factors reduce the possibility that fire records have natural sources on the QTP. Therefore, it is plausible to attribute the fire record to humans.

Fire generally only changes ecosystems in humid climates, which provide enough phytomass to be burned during a dry season. Few studies have examined how fire use by early humans impacted the Tibetan highlands. Picea and Juniperus charcoal of the northeastern highlands have been dated to between 10.0 and 6.5 cal kyr B.P (Kaiser et al., 2007; Lu et al., 2018). Phylogeographic analyses of now disjunct forest relicts with shared haplotypes across the northeastern pastures suggest that post-glacial recolonization resulted in continuous forests that have been fragmented during the Late Glacial and the Holocene (Zhang et al., 2005; Meng et al., 2007). Charcoal and pollen records point to a forest decline (Picea and Betula) after 8 to 6 cal kyr B.P. (Shen et al., 2005; Herzschuh et al., 2006; Cheng et al., 2010; Miehe et al., 2014), at a period that was the most favorable climatic period of the Holocene (Mischke et al., 2005; Chen et al., 2020), with East Himalayan alpine treelines at a position between 300 and 600 m higher around 8 cal kyr B.P. than today (Kramer et al., 2010a; Schickhoff et al., 2016). However, it remains unknown whether hunters used fire, for example, after 7.2 cal kyr B.P. in the Gonghe Basin, as suggested by Wei et al. (2020), or whether fire was the tool of pastoralists to manage their pastures. In the Hengduan Mountains, fire has been recorded since 13 cal kyr B.P., and a decline of Betula pollen between 8.1 and 7.2 cal kyr B.P. occurs synchronously with charcoal peaks (Kramer et al., 2010a, 2010b). Sites in southern Xizang show charcoal peaks between 11.3 and 6 cal kyr B.P (Duo, 2008). or a continuous Holocene fire records (Miehe et al., 2009b; Bird et al., 2014).

Pollen and spores are, like charcoal, ambiguous tools for the detection of human-induced forest-changes. They require knowledge of the regional flora for possible attribution of pollen-types to local plant species, current vegetation distribution, including the impact of livestock, and the ecological indicator values of species. Common pitfalls are the recognition of pollen-types that never have been in the area (e. g., long distance pollen of Pinus or Quercus) or lack of efforts to determine which species in a given area can be assigned to a pollen type in order to be able to determine the pollen values of a species important for the objectives instead of the pollen curve of a more or less meaningless plant family (e.g., Stellera chamaejasme instead of Thymelaeaceae).

In the ‘forest climates’ of the Eastern Highlands (Figs. 4 and 5: space between the green band and two treelines), the current ‘alpine meadows’ are almost certainly a replacement (‘plagioclimax’) of forests. The pollen diagram of Lake Luanhaizi (Fig. 8; Herzschuh et al., 2006) is best suited to demonstrate the interrelationships. The current vegetation around the site is ‘alpine meadow’ (Hou 2001). The climate data show the forest climate, and the endemic Picea crassifolia trees, planted in the court yard of the ‘Haibei Alpine Meadow Ecosystem Research Station’, demonstrate the current potential of trees. The tree-pollen decline in the Mid-Holocene Climatic Optimum is almost certainly not climate-driven, but to be attributed to the decision of humans to change the forest into grasslands. As the pollen record does not show the usual post-fire succession (Epilobium > Populus > Betula), it is plausible that humans suppressed forest recovery because they preferred grassland instead of forest. The synchronous emergence of pollen-types classified as grazing-indicators (e.g., Stellera chamaejasme, Bassecoia hookeri, Polygonum aviculare, Tribulus terrestris) supports this view. However, the knowledge of indicator values of pollen types attributed to humans as developed in Europe (Behre, 1981; Gaillard, 2007), remains poor in the ‘alpine meadows’ (Miehe et al., 2019), and endemic species may have been favoured through selective grazing of the wild herbivores. This change from forests to ‘alpine meadow’ took place between the 10th and 7th millennium before present. If the coincidence of decreased tree-pollen and increased grazing indicator pollen-types is corroborated by synchronous charcoal-, biomarker- and archaeological-records, this change of the forest cover is possibly part of the agro-pastoral transition and the dawn of pastoralism in the highlands. To date, however, there is an interdisciplinary gap in our knowledge of the 5000 years between the palaeo-ecological and the archaeolocical and zoo-archaeological records.

5. Recent changes

The treeline ecotone of all exposures in the eastern highlands displays an upslope gradient of declining tree size (Fig. 9c). This decline in tree size is presumably an effect of global warming enhancing the survival of seedlings and/or decreased grazing pressure. Repeated photography between August 1994 (Fig. 9a) and August 2018 (Fig. 9b) of a site east of Yajiang (4160 m, 30°05′N, 101°24′E) revealed weak changes: The globular-shaped Quercus aquifolioides and Juniperus pingii var. wilsonii have encroached the pastures upslope by approx. 50 m. The oak may attain tree-size while this juniper taxon remains a shrub, even attaining 3 m in height. Another feature to be noted is the obviously recent spreading of conifers on a slope that was once should occupied by oaks (Fig. 9d). The pattern suggests that Picea or Abies are obviously faster in dispersal than Quercus. So far, only the authors' own observations in the field are available; however, it is clear that long-term studies would be necessary to assess this pattern. Since the onset of free-range grazing of livestock, the prevalent pastoral footprint of environmental change and resource decline has been the inhibition of forest regeneration caused by selective browsing of trees and saplings. For 20 years, governmental regulation and/or economic changes, including the exodus from remote highland villages to down-country cities, has decreased livestock numbers in parts of the highlands. The consequent reforestation is obvious; however, it is often seen exclusively as an effect of Anthropocene global warming (Kreutzmann, 2012; Ptackova, 2012; Schickhoff et al., 2022). However, a recent work found a rapid transition from spruce-fir to pine-broadleaf forests in response to disturbances and climate warming on the southeastern QTP, and suggested that both the effects of anthropogenic disturbances and climate on subalpine forests should be considered in adaptive forest management and in projections of future forest changes (Zhang et al., 2023).

6. Conclusions

‘Alpine meadows’ of the Qinghai-Tibet Plateau represent the largest conversion of mountain forests into pastures worldwide. Although the co-occurrence of decreased tree-pollen (e.g. Shen et al., 2005; Herzschuh et al., 2006; Miehe et al., 2014) and increased grazing indicator pollen-types (e.g. Leipe et al., 2014; Miehe et al., 2021) requires corroboration from synchronous charcoal (Kaiser et al., 2007; Lu et al., 2018), biomarker (Callegaro et al., 2018), genetic (Du et al., 2010; Guo et al., 2006; Zhang et al., 2010) and archaeological studies (Chen et al., 2015, 2020), this change of the forest cover is possibly part of the agro–pastoral transition and the dawn of the Anthropocene on the QTP. To date, however, there is an interdisciplinary gap in knowledge of 5000 years between the palaeo-ecological evidence of environmental changes (around 8.5 ka BP), most plausibly explained as a human impact, and the absence of archaeological records during this era. The ‘gap’ thus articulates the discrepancy between the evidence (on the palaeo-ecological side) and the absence (on the archaeological side). This is of course not a judgement of wrong or right, but probably only reflects differences in the density of records. Current changes in the treeline ecotone are likely caused in part by the changing rural economy and the exodus from remote highland villages to down-country cities, which have diminished age-old impacts of summer grazing and pasture management by fire. Consequently, reforestation is obvious, although it is often seen exclusively as an effect of Anthropocene global warming. We recommend that researchers expand existing permanent treeline monitoring assets with unified and transdisciplinary tools, including social sciences for surveys of changes in rangeland management. We believe that more evidence and interdisciplinary collaborations are required to test existing and emerging hypotheses about the treelines of the Anthropocene in High Asia.

Acknowledgements

Georg Miehe acknowledges the enduring support of the German Research Council (DFG) since 1976 and the cooperation with Sichuan University, Yunnan University, and the Institutes of the Chinese Academy of Sciences (CAS) in Kunming, Chengdu, Lanzhou, Xining, and Beijing. Udo Schickhoff is also grateful to the DFG for funding treeline-related research (SCHI 436/14–1). Kangshan Mao acknowledges the National Natural Science Foundation of China (grant numbers U20A2080 and 31622015) and Sichuan University (Institutional Research Fund, 2021SCUNL102; Fundamental Research Funds for the Central Universities, SCU 2022D003). The authors acknowledge the considerable support from anonymous reviewers and editors during the reviews.

Author contributions

G.M. conceived the work and wrote the draft. K.M., S.U.H., J.B, U.S. contribute new thoughts and expand the draft. All authors revised and approved the manuscript. G.M., K.M. finalized the manuscript.

Declaration of competing interest

This manuscript is not under consideration for publication in any other journal, and all authors declared that we have no competing interests.